"""
使用直方图、盒图等检查数据分布及离群点
"""
# 数据可视化，绘制数值属性的直方图，盒图及散点图
def box_plot(df,ylabel=None,new_df=None):
if new_df is None:
plt.figure()
plt.title('Boxplot')
sns.boxplot(y=ylabel,data=df,palette='Set2')
else:
plt.figure(figsize=(14,8))
plt.subplot(121)
plt.title('Boxplot')
sns.boxplot(y=ylabel,data=df,palette='Set1')
plt.subplot(122)
plt.title('new_Boxplot')
sns.boxplot(y=ylabel,data=new_df,palette='Set2')
def hist_plot(df,ylabel=None,bins=10,new_df=None):
if ylabel is None:
print("Please provide ylabel")
return None
if new_df is None:
plt.figure()
plt.title('Histogram')
sns.distplot(a=df[ylabel].dropna(),bins=bins,hist=True,kde=False)
else:
plt.figure(figsize=(14,8))
plt.subplot(121)
plt.title('Histogram')
sns.distplot(a=df[ylabel].dropna(),bins=bins,hist=True,kde=False)
plt.subplot(122)
plt.title('new_Histogram')
sns.distplot(a=new_df[ylabel].dropna(),bins=bins,hist=True,kde=False)
def scatter_plot(df,xlabel=None,ylabel=None,new_df=None):
if xlabel is None or ylabel is None:
print("Please provide xlabel and ylabel")
return None
if new_df is None:
plt.figure()
plt.title('Scatter plot')
sns.scatterplot(x=xlabel,y=ylabel,data=df)
else:
plt.figure(figsize=(14,8))
plt.subplot(121)
plt.title('Scatter plot')
sns.scatterplot(x=xlabel,y=ylabel,data=df)
plt.subplot(122)
plt.title('new Scatter plot')
sns.scatterplot(x=xlabel,y=ylabel,data=new_df)
def viz_pairs(df,ylabel=None,xlabel=None,bins=20,new_df=None):
if ylabel is None or xlabel is None:
print("Please provide xlabel and ylabel")
return None
box_plot(df,ylabel=ylabel,new_df=new_df)
box_plot(df,ylabel=xlabel,new_df=new_df)
hist_plot(df,ylabel=ylabel,bins=bins,new_df=new_df)
hist_plot(df,ylabel=xlabel,bins=bins,new_df=new_df)
scatter_plot(df,xlabel=xlabel,ylabel=ylabel,new_df=new_df)
